Our goal is to develop a computational framework for cell signaling networks, and to demonstrate its effectiveness by developing high-quality models for several pathways of special biological interest. Its purpose is to enable biologists to use mathematical modeling to frame and constrain hypotheses before experimental testing. The need for the system and its design are driven by the hallmarks of cell signaling - complexity and dynamics. Web-like diagrams of cross-talking pathways and information flow currently convey the complexity. Dynamic properties include both transient and permanent changes in network architecture in which pathway components move from one subcellular location to another, components associate in different protein complexes, and the entire repertoire of receptors, transducers and targets and connectivity change as the cell state changes. To mathematically probe signaling processes, we are developing a "pathway modeling database" (SIGMOID) designed to: 1) Capture and organize diverse expert knowledge and data in a manner that computationally mirrors the underlying biochemistry. Pathways are therefore represented as their underlying chemical reactions and molecular associations. It also specifies signal-ontingent spatial compartmentalization of reactants and products, and provides for biologically appropriate dependence of reactions on the molecular state of reactants (phosphorylation, methylation, GDP binding, etc). 2) Provide automated generation of computational models based on the stored reaction network architecture that is coupled with an expandable suite of modeling programs. 3) Ingest and store results from computational simulations and provide tools for comparing, visualizing and mining simulation results.